Comparing SVM and Neural Networks’ performance in Face Detection
Shivam Bhatia, Uditanshu Tomar, Aryan Vikas Jain · 2021 International Conference on Intelligent Technologies (CONIT) · 2021
There exists an abundance of literature where researchers have attempted to improve security using face detection and recognition. Having studied the diverse models proposed in these works, we have come to appreciate the insightful nature of the current research on security and privacy. Although the current research works are commendable, we argue that a thorough examination of the methods can help address the pitfalls to come forward with even more advanced models. The application of computer vision in augmenting security and privacy captivated our attention the most. Thus, we draw comparisons between Support Vector Machines (SVMs) and Neural Networks (NNs) for face recognition from images. The frequent usage of SVM and NN over the years in face detection applications prompted us to choose these two approaches. Keeping data augmentation and HOG as the feature extraction method, we compare the models developed to find the best approach for facial recognition. We used the combination of F1 Score, Accuracy, Precision, and Recall to compare both models.